Pathologist Interrater Reliability and Clinical Implications of Elevated Donor-Derived Cell-Free DNA beyond Heart Transplant Rejection.

Abstract

BACKGROUND There is significant variability amongst pathologists in the histopathological interpretation of the endomyocardial biopsy (EMB) for acute cellular rejection (ACR) and assessment of variability in the interpretation of antibody-mediated rejection (AMR) has not been reported. In contemporary practice, the strategy of allograft surveillance with donor-derived cell-free DNA (dd-cfDNA) as compared to EMB has not been compared with a focus on long-term clinical outcomes beyond acute rejection (AR). METHODS The Genomic Research Alliance for Transplantation (GRAfT) is a multicenter, prospective cohort study that enrolled patients from 2015 to 2020. The center pathologist read was compared to two blinded core cardiac pathologists. ACR and AMR were graded based on the International Society for Heart and Lung Transplantation (ISHLT) criteria. Weighted Cohen's kappa (κ) was used to evaluate interrater reliability between the center and core reads. To assess long-term outcomes, we evaluated a composite of AR, allograft dysfunction, and mortality within 1 year. RESULTS The study included 94 patients (median age 55 years [IQR 45, 62]), 30% female, 41% Black race) with a total of 429 EMBs and paired dd-cfDNA measures. The concordance rate between center and core pathologists was 77% for ACR (95%CI: 66% - 89%) and 63% for AMR (95%CI: 53% - 74%). 46 patients had an elevation in dd-cfDNA without AR by EMB. The median dd-cfDNA was 0.49% (IQR: 0.35, 1.01) and subsequent AR, allograft dysfunction, or mortality occurred in 59% of these patients at 1 year. In patients with AR by EMB and negative dd-cfDNA (n=5) the composite outcome occurred in 20% of patients at 1 year. At baseline, the positive likelihood ratio (LR+) of dd-cfDNA to detect AR by the center pathologist was 3.74 (95% CI 3.01 - 4.64) and core pathologist was 2.59 (95%CI: 1.95 - 3.45). If the composite outcome was included as a true positive, the LR+ of dd-cfDNA improved to 9.82 (95%CI: 7.04, 13.69) and7.63 (95% CI: 5.61, 10.38) at 1-year, respectively. CONCLUSIONS Pathologists interrater reliability is limited in both ACR and AMR. The positive LR of dd-cfDNA when compared to traditional histopathology is limited, but when longitudinal clinical outcomes are included to assess diagnostic performance, the LR+ improves significantly. The value of dd-cfDNA extends beyond the diagnosis of AR to include other clinically meaningful outcomes for patients after heart transplant.